- Email: [email protected]
Glacial tectonics: a deeper perspective
Quaternary Science Reviews 19 (2000) 1391}1398
Glacial tectonics: a deeper perspective Robert M. Thorson* Department of Geology and Geophysics, University of Connecticut, 345 Mansxeld Road (U-45), Storrs, CT 06269, USA
Abstract The upper 5}10 km of the lithosphere is sensitive to slight changes ((0.1 MPa) in local stress caused by di!erential loading, #uid #ow, the mechanical transfer of strain between faults, and viscoelastic relaxation in the aesthenosphere. Lithospheric stresses induced by mass and #uid transfers associated with Quaternary ice sheets a!ected the tectonic regimes of stable cratons and active plate margins. In the latter case, it is di$cult to di!erentiate glacially induced fault displacements from nonglacial ones, particularly if residual glacial stresses are considered. Glaciotectonics, a sub-subdiscipline within Quaternary geology is historically focussed on reconstructing past glacier regimes and, by de"nition, does not include these e!ects. The term `glacial tectonicsa is hereby suggested for investigations focussed on the past and continuing in#uences of ice sheets on contemporary tectonics. 2000 Elsevier Science Ltd. All rights reserved.
1. Introduction Presently, there is a conceptual shift in the geosciences away from increasing specialization, towards more integrative problems at global scales, a shift embodied by the phrase `Earth System Sciencea (Kump et al., 1999). Simultaneously, technologically driven advances in instrumental and computational techniques are permitting earth scientists to identify and explain synoptic variations in topography, seismicity, and ambient crustal stress that were not recognizable a decade ago. The recent paper by Peltzer et al. (1996) is a case in point; they used synthetic aperture radar (SAR) interferometry to explain the disappearance of transient topographic anomalies at the centimeter scale resulting from the 1992 Landers (California, USA) earthquake, anomalies erased by the in-migration of #uids to 4 km depth. As these conceptual and technical trends continue, the seismotectonic e!ects of past #uctuations of Quaternary ice sheets are becoming easier to notice and explain. As a result, the role of glaciers in earth deformation is being recognized at a range of spatial scales: from deformation within the ice itself to the reactivation of tectonic faults in response to crustal unloading and viscoelastic relaxation of isostatic anomalies. Broader recognition of the multiple
* Tel.: 860-486-1396; fax: 860-486-1383. E-mail address: [email protected] (R.M. Thorson).
e!ects of glacial mass transfers is blurring the distinction between the study of tectonics, per se, and the study of `glaciotectonicsa, which, historically, has been primarily concerned with deformed glacial deposits. In this paper I show how the physical coupling between glaciation and crustal deformation extends far beyond the decollement between ice and its substrate (i.e. beyond the scope of glaciotectonics), and instead operates up to crustal scales. I begin by reviewing recent research illustrating the sensitivity of the earth's crust to stress di!erences far smaller than those associated with ice sheets. I then introduce examples of glacially induced, crustal scale deformation from a passive cratonic setting (Fennoscandia) and an active continental margin (Cascadian subduction zone). Next, I explore the basic mechanisms of glacially induced tectonics * e!ective vertical stress, membrane #exure, and traction at the base of the crust * and how each of these mechanisms is modi"ed by the contrasting tectonic domains associated with compressive, extensional, and transform strain. Finally, I revisit the question of the scope of glaciotectonics.
2. Sensitivity of the Earth's crust The ambient seismicity in many regions is extremely sensitive to small changes in stress, regardless of cause. For example: Rydelek and Sacks (1999) recently demonstrated that seemingly trivial changes in the con"ning
0277-3791/00/$ - see front matter 2000 Elsevier Science Ltd. All rights reserved. PII: S 0 2 7 7 - 3 7 9 1 ( 0 0 ) 0 0 0 6 8 - 8
1392
R.M. Thorson / Quaternary Science Reviews 19 (2000) 1391}1398
stress (0.1 MPa) are su$cient to nucleate earthquakes; King et al. (1994) proved that fault displacement in one locality can transfer stress to other faults at continental scales; and Pollitz et al. (1998) implicated the aesthenosphere in the transfer of stress across the North Paci"c Ocean from the Aleutian Islands to California. The role of crustal pore pressure in tectonic processes has also been recently clari"ed. For example: Rojstaczer and Wolf (1992) documented a tenfold increase in baseflow stream discharge following the 1989 Loma Prieta (California, M 7.1) over a radius up to 50 km from the epicenter; Episodic movements on great transform faults (Byerlee, 1993; Sample and Reid, 1998) are toggled by pore pressure variations; Thrust faults within accretionary wedges at continental margins (Bolton et al., 1999) are regulated by hydro-mechanical (valving) processes which partition an otherwise continuous process (subduction) into discrete episodes of deformation. Decollemonts would be mechanically impossible without high pore pressures. Arti"cial triggering of earthquakes by human activities, now well established, works primarily through the e!ects of #uid pressure. For example, the injection of petroleum brines, waste-water, and experimental tracers have initiated brittle faulting to a depth of 9 km (Jost et al., 1998). Conversely, the withdrawals of #uids in#uence seismicity to a minimum depth of 4 km (Gomberg and Wolf, 1999; Segall, 1989). The so-called `delayeda response of reservoir seismicity has been shown to be a consequence of the migration of #uid pressure towards potential asperities on faults, with rupture occurring when a threshold pressure is reached. Dry e!ects such as crustal loading by water-supply reservoirs (Simpson et al., 1988; Mandal et al., 1998), the extraction of mass from large open-pit mines, and underground explosions are generally less important than #uid e!ects. Stress changes caused by human activities are small with respect to those of ice sheets, present or past, which also precipitate and dampen earthquake processes. For example, Johnston (1987) calculated that the East Antarctic Ice Sheet was responsible for a loading stress reaching 59 MPa, although the change in e!ective stress is much lower. Glaciers, as agents of denudation, are also responsible for permanent mass transfers from highlands to depositional basins, but at a much slower rate than ice loads. The closest analogue to a glacially induced seismicity from #uids was the damaging Koyna (India) earthquake of December 10, 1967 the largest known reservoir earthquake (Mandal et al., 1998). Its main shock was a strike-slip event with a magnitude of 6.2 (Simpson et al., 1988) and a focal depth of less than 5 km, but aftershocks extended to at least 15 km depth in a zone nearly 10 km wide and 30 km long. Carefully instrumented variations in aftershock micro-seismicity indicate that the state of stress in the crust in Koyna was controlled by inter-annual variability between
wet and dry years as well as the water level in the reservoir. Earthquake nucleation processes similar to those beneath the Koyna reservoir * dry loading, poroelastic volume changes, #uid transfers * took place on a much larger scale in the vicinity of former Quaternary Ice Sheets. Although the direct loading and pore pressure e!ects are most easily understood, the growth and decay of ice sheets were also associated with crustal #exure and aesthenospheric relaxation. Although recognized as important for more than century, the tectonic coupling between ice sheets (the water component of the lithosphere) and the crust (the silicate component) is seldom considered in geologically based reviews of the seismotectonic literature (Yeats et al., 1997; Keller and Pinter, 1996).
3. Contrasting examples Postglacial fault scarps in northern Fennoscandia are the dramatic surface expressions of faults penetrating the Baltic shield to a depth of about 40 km (Arvidsson, 1996). The most conspicuous, the Parvie Fault in northern Sweden, with a vertical displacement of 10 m and a rupture length of 150 km (Lagerback, 1992; Muir-Wood, 1988) may have generated a deglacial earthquake with an extraordinary magnitude (Mw"8.2). Similar, albeit smaller, faults are known for the Canadian and Siberian shields (e.g. Dyke et al., 1991), in tectonic settings that requires the `externally imposed failure conditionsa (Johnston, 1996) caused by glacial loading. Although the Parvie Fault, as well as their counterparts on other shields, was formed during deglacial mass transfers, it remains somewhat active today, as evidenced by continuing micro-seismicity. Ironically, modern seismicity has a glacial heritage. Building on previous models (Hasegawa and Basham, 1989; James and Morgan, 1990; Johnston, 1987; MoK rner, 1980; Muir-Wood, 1988; Stewart and Hancock, 1994) I applied the idea of crustal-scale glaciotectonic deformation to an active plate margin in the Cascadia Subduction Zone, arguing that the tectonic processes in the fore-arc basin of the Puget Lowland were, and continue to be, compromised by the mass and #uid transfers associated with Quaternary glaciation (Thorson, 1996). More speci"cally, I predicted that a much simpler, preglacial ambient tectonic regime has become partitioned into four discrete time domains, all of which occur sequentially during each glacial cycle: E A phase of normal background stress in which the direct and delayed e!ects of glacial mass transfers are insigni"cant (normal Coulomb failure). E A phase of mass loading/crustal strengthening in which the combination of ice thickening and isostatic
R.M. Thorson / Quaternary Science Reviews 19 (2000) 1391}1398
out#ow reduced the deviatoric stresses, contributing to stability. E A phase of unloading/weakening during early postglacial time associated with signi"cant fault displacements caused by #uid overpressuring and enhanced traction at the base of the crust due to isostatic in#ow. E A recovery phase in which ambient tectonic stresses, supposedly `eraseda during the deglacial phase, reaccumulated to background values. An attempt to test this hypothesis with "eld data (Holocene fault displacements) was encouraging, but inconclusive. Since publication of that paper, the evidence for linking glacial unloading with a transient phase of intense deglacial tectonism has been strengthened. Nearly 40 years ago, Easterbrook (1963) published evidence for an enigmatic emergence and re-submergence of a crustal block near Deming, in northwestern Puget Lowland, during glaciomarine conditions about 12 ka. Owing to the magnitude of the required oscillation (30}70 m), the interpretation of these features remains controversial. Recently, Dragovich et al. (1997) linked these vertical motion anomalies (as well as a concentration of bedrock landslides) to the leading edge of the upper plate of a thrust fault exhibiting both Neogene o!set and modern seismic activity. Emergence was likely caused by at least 30 m of localized tectonic uplift during the most rapid phase of deglaciation, an interval when isostatic uplift took place at a rate between 10 and 30 cm yr\ (Dethier et al., 1995), and when the rapid northward #ow of a low-viscosity mantle was in phase with the plate convergence direction (10}10 poise; James et al., 2000). Tectonically, this uplift can be interpreted as a structural wedge caught between the leading edge of a Quaternary thrust fault and an antithetic reverse fault behind it. The subsequent re-submergence is more enigmatic, but may have been caused by normal faulting and backtilting associated with a subsequent phase of crustal extension associated with continued growth of the uplift dome (Muir-Wood, 2000). Curiously, two other elevated anomalies in uplift pro"le * Mt. Vernon (Dethier et al., 1995) and Alki Point (Thorson, 1996), also lie on the upper plate of thrusts (Johnson et al., 1999). The amplitudes of the topographic anomalies ('30 m for Deming on the McCauley Creek Thrust, ca. 22 m for Mt. Vernon on the Devil's Mountain Fault, and ca. 9 m for Strand A of the Seattle Fault) scale with fomer ice thickness and with the magnitude and rate of isostatic uplift.
4. Glacial tectonics The magnitude of the deglacial uplifts involved in Puget Lowland and the intimate connection between
1393
pre-existing tectonic structures and glacially induced deformation, blurs the distinction between tectonics and glacially induced tectonics. Strictly speaking, the glacier itself is a tectonic phenomenon. The base of each ice sheet is a rapidly propagating, gravity-driven detachment fault, taking place entirely within the lithosphere. The loss of ice by melting is analogous to un-roo"ng by denudation 2The glacier sole is essentially a slickensided decollement, smeared with fault gouge. The rigid lower plate (the crust) undergoes elastic #exure and changing deviatoric stresses caused directly by the load, and indirectly by the transient decay of the isostatic anomaly and excess pore #uid pressure. (Thorson, 1996, p. 1187) What separates glacial tectonics from `normala tectonics is less the physical processes involved or the scope of their in#uence, but the time scale.
5. Glacial tectonic mechanisms Kinematically, there are three principal directions in which the forces associated with a change in glacier mass can in#uence the state of stress of the lithosphere: The stresses can be imposed: vertically through changes in e!ective stress; horizontally by imparting a shear traction at the top of the crust (conventional glaciotectonics) or at its base (horizontal component of the viscoelastic response in the aesthenosphere); or three dimensionally with respect to volumetric strain and #exure. All of these e!ects, treated separately below, take place simultaneously, and can overlap in time (Fig. 1). 5.1. Ewective vertical stress Here we consider three limiting cases. In the simplest limiting case, pore pressure e!ects are absent, allowing the growth of an ice sheet (of in"nite width) to create a direct, instantaneous, lithostatic increase in the vertical stress: *p "o g h , T where *p is the change in vertical stress, o is the density of ice, and h is the ice thickness. This e!ect is equivalent to the `rapida reservoir response in induced seismicity directly beneath impoundments. Owing to the Poisson e!ect, however, the increased vertical stress also causes a proportional (but lower) increase in the horizontal stress, or *p , reducing the potential deviatoric (i.e. tec& tonic) stress. The importance of loading diminishes with depth as the stress ratio between pre and post-loaded conditions declines. The principal e!ects are elastic and instantaneous.
1394
R.M. Thorson / Quaternary Science Reviews 19 (2000) 1391}1398
Fig. 1. Schematic diagram showing glaciotectonic domains and processes. (a) Three fundamental processes associated with changes in glacier distribution; elastic e!ects, vertical stresses, and shear. (b) Idealized physical model for changes in vertical stress during loading. The change in e!ective stress is determined by changes in pore pressure (k) and vertical load (*p ), and are reversed and generally more rapid during unloading. Crustal 4 permeability (k) after Manning and Ingebritsen (1999). Change in vertical load with depth is idealized. Top part of the physical model has no scale; the thickness of layers is exaggerated to show di!erential responses.
In a second limiting case, a wet-based ice sheet is saturated with #uid water to near its surface, and the water has unlimited access to the base in an isotropic crust. In this circumstance, the change in e!ective vertical stress (*p ) is zero because the increased #uid pressure # (*k) fully compensates the increased vertical load (*p ). Alternatively *p "0"(*p !k). # This limiting case is realistic for ice-loaded shield areas well inside the glacial margin, where fractures in strong rocks to a depth of several kilometers are open to the surface, and where the high fracture permeability does not restrict the #ow of #uids. In these circumstances, changes in the e!ective vertical stresses remain close to zero because the #uid pressure instantaneously mirrors the changes in load. Changes in horizontal stresses through the Poisson e!ect remain important. In cases where the potentiometric surface of #uid water within the ice sheet falls to a level appreciably below the glacier surface, then the e!ective vertical stress is raised by the proportional decrease in #uid pressure. Conversely, during glacier surges, the potentiometric pressure from meltwater (o"1.0 g/cm) may exceed the added vertical load of the ice (o"0.9 g/cm), causing a transient reduction in the e!ective vertical stress during a time of rapid mass transfer. Thus, micro-seismicity associated
with glacier surges (Kamb et al., 1985) likely takes place in the bedrock, as well as within the ice. The permeability of the Earth's crust, whether sedimentary basins, fractured crystalline basement, and volcanic rocks, decreases systematically with depth, owing to increased con"ning pressure and the closure of fractures (Fig. 1). Based on studies of metamorphic #uids and heat transport (Manning and Ingebritsen, 1999), the permeability (k) in SK units of `ma varies with crustal depth (z, km) in a simple, log}log relationship spanning at least 10 orders of magnitude: Log k"!14}3.2 log z. Log k for sedimentary basins to a depth of 7 km ranges from !12 to !16 m\. Volcanic rocks and crystalline basement are typically one-to-two order of magnitudes less permeable. Although conceptually straightforward, de"ning the boundary between meteoric and connate water is di$cult in actual practice. The relationship between permeability and depth is least reliable in the shallow crust owing to greater variation in horizontal stress and to lithological heterogeneity. At some depth, fracture aperture is small enough to limit the permeability, forcing glacially induced pressure changes to take place gradually. The depth at which this transition takes place is not known, but likely ranges widely from about 0.5 km to more than several km. This time-dependent migration of #uids in the shallow crust
R.M. Thorson / Quaternary Science Reviews 19 (2000) 1391}1398
(Peltzer et al., 1996) is similar to the `delayeda e!ect for reservoir seismicity (Simpson et al., 1988) and injection experiments (Odonne et al., 1999), which takes place over time scales of weeks to years. In the glacially induced version of this scenario, changes in #uid pressure of `meteoric watera in the open, but low-permeability crust, will lag each instantaneous increment of load. This will cause an instantaneous rise in the e!ective vertical stress that will decay as the pressure re-equilibrates. Non-linear changes in the permeability may occur during re-equilibration, possibly through the geochemical e!ects (stresscorrosion) of in"ltrating water (Kisslinger, 1976). At greater depth, however, where poroelastic e!ects (Segall, 1989) predominate, connate #uid plays a critical role in supporting the crust. A glacially induced increase in vertical stress (*p ) will be fully compensated by an instan taneous rise in #uid pressure, one that will gradually bleed o! through #uid out-migration. This will cause a transient phase during which an initially weakened crust will strengthen, an e!ect opposite that in the next-higher layer, where the initially strengthened crust will weaken. At even greater depths, #uids are not free to migrate upward, the proportional change in hydrostatic pressure diminishes, and anisotropy plays a critical, and largely unknown, role in modulating the e!ects of the ice load. Near the edge of the glacier, however, the e!ects will be more complicated. Disregarding #uid e!ects, the di!erential weighting on either side of a glacier margin will be felt immediately as the crust is elastically compressed below the load, with no compression beyond it. When #uid e!ects are considered, the di!erential stresses are greater because the glacier imparts a new hydraulic gradient on the system. Seepage pressures, particularly near the ice margin, may also in#uence the ambient stress. Membrane #exure of the crust during isostatic compensation will cause radial compression and extension beneath the ice load and in the forebulge domain, respectively, deformation that will a!ect crustal permeability. 5.2. Horizontal traction; top and bottom Drag forces at the glacier}bed interface are controlled by the e!ective overburden stress (ice thickness reduced by #uid pressure), the viscoelastic properties of the ice, and the frictional properties (Mohr}Coulomb) of the boundary. Three limiting cases can be visualized. In the "rst case, thick ice is frozen to a #at, crystalline bed; horizontally #owing ice and the stationary bed form an ideal simple shear couple limited only by the slow rate of ice deformation; in this case, the shear on the rock is trivial. In the second case, the ice moves continuously above a very weak, deforming bed, which fully compensates the shear stresses. As before, negligible shear is propagated downward to the lithosphere. In the third case the ice is frozen (or frictionally bonded) to its bed near the margin, with groundwater out#ow taking place beneath the ice}bed
1395
interface in strata where the permeability is limiting. In this circumstance, the frictional bond between the glacier and its bed (whether rock or unconsolidated sediment) is stronger than in deeper materials, even that of jointed bedrock. The traction associated with the advancing ice can thus be transferred downward into the crust, with displacement along faults, both brittle or ductile, taking place to a depth of up to several hundred meters, and involving intact slabs of the lithosphere at the kilometer scale (Aber, 1988). The depth and behavior of such e!ects are controlled primarily by the strain rates being applied. Thrusting at the ice margin, which leads to the familiar push moraines and imbricate ridges, is conceptually similar to the behavior at accretionary wedges of subduction zones. A second type of horizonally}directed shear takes place at the base of the rigid crust through the e!ect of horizontal #ow within the asthenosphere (James and Morgan, 1990). Traditionally, studies of glacioisostasy have focussed on the vertical component of motion and on crustal #exure (MoK rner, 1980), but the horizontal #ow between the forebulge and the subglacial depression, can be quite rapid, up to a meter per year. Recent geodetic investigations in the western US con"rm that viscoelastic relaxation in the aesthenosphere after major earthquakes (Rydelek and Sacks, 1999) raises the strain along other, distant faults. Similarly, horizontal isostatic motions during loading and unloading must enhance or diminish the horizontal components of regional stress (Muir-Wood, 2000). 5.3. Three-dimensional yexure Mechanical loading is has been reviewed by Turcotte and Schubert (1982). Slab loads, line loads, point loads, end loads, and embedded loads all create #exures of the lithosphere, the size and shape of which are governed largely by the e!ective rigidity of the crust, which, is sensitive to the third power of membrane thickness. The physics of the process are independent of the composition of the load, whether glacier ice, crystallized lava, or a sedimentary accumulation. Hence, the familiar `moatforebulgea pro"les of volcanic islands, foreland basins, and glacial margins are similar. Importantly, membrane #exure includes zones of convexity where the crust is radially dilated, and zones of concavity where it is radially compressed. Radial and tangential extension on the forebulge, and radial compression beneath the ice sheets took place during the loading hemicycle; the reverse took place, and with higher strain rates, during deglaciation, a process that continues today. 6. Complications 6.1. Ambient crustal stress In the previous discussion, I made the simplifying assumption that the ambient lithospheric stress was
1396
R.M. Thorson / Quaternary Science Reviews 19 (2000) 1391}1398
isotropic. Lithospheric stress, however, varies continuously and heterogeneously in response to plate motion, topographic loading, exhumation, and the mass transfer of water during ice ages. Generally speaking, the state of crustal stress can be classi"ed into one of three categories: In a compressive, or thrust regime, the principal stress (p ) is horizontal and the minimum stress (p ) is vertical. Increased e!ective vertical stresses stabilizes the system by reducing the deviatoric stress. Conversely, a reduction in e!ective vertical stress pushes the system towards instability. Fluid injection often produces earthquakes with focal solutions consistent with at least a component of reverse faulting (Jost et al., 1998) because the e!ect of #uid injection is to raise the pore pressure, which reduces the e!ective vertical stress. In an extensional, normal fault regime, p is vertical and p is horizontal. Increas ing the e!ective vertical stress helps destabilize the system, whereas reducing the vertical stress stabilizes it. Fluid extraction (Gomberg and Wolf, 1999) increases vertical stress by reducing the pore pressure, and is thus often associated with normal faulting (Amelung et al., 1999). In a strike slip regime, where p and p are both horizontal, an increase in the e!ective stress stabilizes the system through the Poisson e!ect; pore pressure is especially important in regulating strike slip faults. Although there are broad tectonic domains where the ambient stress is relatively uniform, local stresses are often highly heterogeneous at a variety of scales (e.g. Mueller et al., 1999). Additionally, most faults have an oblique component of motion involving two of the idealized regimes above. 6.2. Glacial history The tectonic e!ects of Quaternary ice sheets are unique because their e!ective loads can change rapidly, more than several orders of magnitude faster than the processes of mountain building and erosion. Although the stresses are lower than for lithospheric forces, the stress gradients and strain rates can be signi"cantly higher, causing brittle rupture where ductile failure might have otherwise occurred. Also, unloading is usually much more rapid than loading, reversing many of the e!ects described above. Importantly, the e!ects of loading and unloading work in opposite directions, allowing the possibility for glaciers to trigger episodes of rupture regardless of ambient stress. They push and pull. 6.3. Coupled ewects Given the three basic directions of applied stress, the three fundamental stress domains, the heterogeneity of local stress within the crust at a variety of scales, and the asymmetric history of ice sheet loading and unloading, glaciotectonic stresses will be complex. Additionally, the e!ects can either amplify or diminish one another. For example:
E The creation of a forebulge may contribute to radial jointing, which may increase crustal permeability, which may decrease the pore pressure beneath the ice margin, which may steepen the ice pro"le, which may cause even greater #exure, etc. E Aesthenospheric out#ow caused by ice loading may enhance or diminish stresses of tectonic origin, depending on the ambient direction of plate convergence. For example, deglaciation of the Cordilleran Ice Sheet near Kodiak Island in southern Alaska may have enhanced and reduced the rate of plate convergence on the southern and northern #ank of the former ice sheet, respectively. E The rate of de-weighting caused by ablation may exceed the rate at which excess pore pressure can drain towards the surface, leaving the crust weaker than it would be under normal conditions. In thrust and strike-slip domains, this may lead to a deglacial interval of heightened seismicity. E Fault displacement triggered by a rise in pore pressure may be a self-limiting phenomenon through the e!ect of stress dilatancy, in which crustal porosity is increased through microfracturing. E Above the onset of poroelastic e!ects the crust (excluding normal fault regimes) is instantaneously strengthened by vertical loading, then weakened back to initial conditions by the in-migration of #uids. The opposite happens where poroelastic e!ects predominate; the crust is weakened, then strengthened. Both of these e!ects can take place simultaneously in di!erent superimposed crustal layers, concentrating stress in subhorizontal boundaries between these zones. E The permeability of the shallow crust is highly variable, hence the e!ects of changing glacial loads will be felt di!erentially; delayed pressure e!ects will continue to in#uence local stresses until re-equilibration, long after the ice has disappeared.
7. Conclusion The tectonic setting of passive continental margins is usually characterized by a thick granitic crust, greater uniformity in ambient stress, and low strain rates (10\ yr\). In such a setting * for example in Fennoscandia (Arvidsson, 1996), the British Isles (Sissons and Cornish, 1982), and the North American Craton (Adams and Bell, 1991; Andrews, 1991; Oliver et al., 1970) * glacially induced stresses are superimposed on the regional "eld, but may actually be the dominant factor controlling seismicity and crustal motion at the time scale of 10}10 yr (MoK rner, 1991). In contrast, the tectonic setting of active continental margins is characterized by high strain rates (10\ yr\), crustal-scale anisotropy, and the dynamic e!ects of plate #exure and buoyancy. In this type of setting * for example, in Cascadia (Mathews
R.M. Thorson / Quaternary Science Reviews 19 (2000) 1391}1398
et al., 1970; Clague, 1983; Thorson, 1996), glacially induced stresses and subduction work simultaneously at Quaternary time scales. In the former case of passive margins, ice sheets modify an existing tectonic regime. In the latter case, they must be considered part of it. Glaciotectonics, as conventionally de"ned, `2involves structural deformations of the upper horizon of the lithosphere caused by glacial stressesa (Van der Wateren, 1995). This de"nition speci"cally excludes deformation within the ice itself, and crustal-scale e!ects because the primary research objective of glaciotectonics, is the reconstruction of past glacier regimes from "eld observations, especially those associated with deformation of unconsolidated materials at and below the glacier sole, and with marginal thrusting at scales of 10\}10 m. The academic territory of glaciotectonics is clearly de"ned as subdiscipline within Quaternary geology. I recommend continued use of the term in this narrowly restricted sense. A new term, however is needed, one that encompasses the deformation in the shallow crust caused by ice}bed traction, and the crustal-scale deformation described in this review. For now I suggest the following candidate terms, arranged in order of increased simplicity: glacio-seismotectonics, glacio-seismicity, glacially induced tectonics, glacial tectonics, or simply tectonics. Although tempted to adopt the last candidate for its simplicity, the term `glacial tectonicsa seems an acceptable compromise between focus and breadth. The actual practice of glacial tectonics has a long pedigree. I recommend adoption of this term to help focus an important, coalescing subdiscipline within geology, one that straddles the boundary between glaciology and tectonics.
Acknowledgements Jeanne Sauber, Iain Stewart and Jim Rose are thanked for reviews that considereably improved the paper. Earlier discussions with Sam Johnson, Jean Crespi and Tom Torgersen were also helpful.
References Aber, J.S., 1988. Bibliography of glaciotectonic references. In: Croot, D.G. (Ed.), Glaciotectonics: Forms and Processes. A.A. Balkema, Rotterdam, pp. 195}210. Adams, J., Bell, J.S., 1991. Crustal stresses in Canada. In: Slemmons, D.B., Engdahl, E.R., Zoback, M.D., Blackwell, D.D. (Eds.), Neotectonics of North America. Geological Society of America Decade Map Volume, Boulder, Colorado, pp. 367}386. Amelung, F., Galloway, D.L., Bell, J.W., Zebker, H.A., Laczniak, R.J., 1999. Sensing the ups and downs of Las Vegas: InSAR reveals structural control of land subsidence and aquifer-system deformation. Geology 27, 483}486. Andrews, J.T., 1991. Late Quaternary glacial isostatic recovery of North America, Greenland, and Iceland: a neotectonics perspective.
1397
In: Slemmons, D.B., Engdahl, E.R., Zoback, M.D., Blackwell, D.D. (Eds.), Neotectonics of North America. Geological Society of America Decade Map Volume, Boulder, Colorado, pp. 473}486. Arvidsson, R., 1996. Fennoscandian earthquakes: Whole crustal rupturing related to postglacial rebound. Science 274, 744}746. Bolton, A.J., Clennell, M.B., Maltman, A.J., 1999. Nonlinear stress dependence of permeability: a mechanism for episodic #uid #ow in accretionary wedges. Geology 27, 239}242. Byerlee, J.D., 1993. Model for episodic #ow of high-pressure water in fault zones before earthquakes. Geology 21, 303}306. Clague, J.J., 1983. Glacio-isostatic e!ects of the Cordilleran ice sheet, British Colombia. In: Smith, D.E., Dawson, A.G. (Eds.), Shorelines and Isostasy. Academic Press, London, pp. 285}319. Dethier, D.P., Pessl Jr., F., Keuler, R.F., Balzarini, M.A., Pevear, D.R., 1995. Late Wisconsinan glaciomarine deposition and isostatic rebound, northern Puget Lowland, Washington. Geological Society of America Bulletin 107, 1288}1303. Dragovich., J.D., Zollweg, J.E., Qamar, A.I., Norman, D.K., 1997. The Macaulay Creek Thrust, the 1990 5.2-magnitude Deming Earthquake, and Quaternary geologic anomalies in the Deming Area, western Whatcom County, Washington * Cause and E!ects? Washington Geology 25, 15}26. Dyke, A.S., Morris, T.F., Green, D.E.C., 1991. Postglacial tectonic and sea level history of the central Canadian Arctic. Geological Survey of Canada Bulletin 397, 56 pp. Easterbrook, D.J., 1963. Late Pleistocene glacial events and relative sea-level changes in the northern Puget Lowland, Washington. Geological Society of America Bulletin 74, 1465}1483. Gomberg, J., Wolf, L., 1999. Possible cause for an improbable earthquake: the 1997 Mw 4.9 southern Alabama earthquake and hydrocarbon recovery. Geology 27, 367}370. Hasegawa, H.S., Basham, P.W., 1989. Spatial correlation between seismicity and postglacial rebound in eastern Canada. In: Gregersen, S., Basham, P.W. (Eds.), Earthquakes of North Atlantic Passive Margins. Kluwer Academic Publishers, Dordrecht, pp. 483}500. James, T.S., Clague, J.J., Wang, K., Hutchinson, I., 2000. Postglacial rebound at the northern Cascadia Subduction Zone. Journal of Quaternary Science Reviews, this issue. James, T.S., Morgan, W.J., 1990. Horizontal motions due to postglacial rebound: Geophysical Research Letters 17, 957}960. Johnson, S.Y., Dadisman, S.V., Childs, J.R., Stanley, W.D., 1999. Active tectonics of the Seattle Fault and central Puget Sound, Washington * implications for earthquake hazards. Geological Society of America Bulletin 111, 1042}1053. Johnston, A.C., 1987. Suppression of earthquakes by large continental ice sheets. Nature 3309, 467}469. Johnston, A.C., 1996. A wave in the earth. Science 274, 735}736. Jost, M.L., Bu{elberg, T., Jost, O., Harjes, H.-P., 1998. Source parameters of injection-induced microearthquakes at 9 km depth at the KTB deep drilling site, Germany. Bulletin of the Seismological Society of America 88, 815}832. Kamb, B., Raymond, C.F., Harrison, W.D., Engelhardt, H.F., Echelmeyer, K.A., Humphrey, N.F., Brigman, M.M., Pfe!er, T., 1985. Glacier surge mechanism: 1982}1983 surge of Variegated Glacier, Alaska. Science 227, 469}479. Keller, E.A., Pinter, N., 1996. Active Tectonics. Prentice-Hall, Upper Saddle River, NJ, 338 pp. King, G.C.P., Setin, R.S., Lin, J., 1994. Static stress changes and the triggering of earthquakes. Bulletin of the Seismological Society of America 84, 935}953. Kisslinger, C., 1976. A review of theories of mechanisms of induced seismicity. Engineering Geology 10, 85}98. Kump, L.R., Kasting, T.F., Crane, R.G., 1999. The Earth System. Prentice-Hall, Upper Saddle River NJ, 351 pp. Lagerback, R., 1992. Dating of Quaternary faulting in northern Sweden. Journal of the Geological Society of London 149, 285}292.
1398
R.M. Thorson / Quaternary Science Reviews 19 (2000) 1391}1398
Mandal, P., Rastogi, B.K., Sarma, C.S.P., 1998. Source parameters of Koyna earthquakes, India. Bulletin of the Seismological Society of America 88, 833}842. Manning, C.E., Ingebritsen, S.E., 1999. Permeability of the continental crust: implications of geothermal data and metamorphic systems. Reviews of Geophysics 37, 127}150. Mathews, W.H., Fyles, J.G., Nasmith, H.W., 1970. Postglacial crustal movements in southwestern British Columbia and adjacent Washington State. Canadian Journal of Earth Sciences 7, 690}702. MoK rner, N.-A., 1980. Earth Rheology, Isostasy and Eustasy. Wiley, New York, 599 pp. MoK rner, N.-A., 1991. Intense earthquakes and seismotectonics as a function of glacial Isostasy. Tectonophysics 188, 407}410. Mueller, K., Champion, J., Guccione, G., Kelson, K., 1999. Fault slip rates in the modern New Madrid Seismic Zone. Science 286, 1135}1138. Muir-Wood, R., 1988. Extraordinary deglaciation reverse faulting in northern Fennoscandia. In: Gregersen, S., Basham, P.W. (Eds.), Earthquakes at North Atlantic Passive Margins: Neotectonics and Postglacial Rebound. Kluwer Academic Publishers, Dordrecht, pp. 141}173. Muir-Wood, R., 2000. Deglaciation Seismotectonics: a principal in#uence on intraplate seismogenesis at high latitudes? Journal of Quaternary Science Reviews, this issue. Odonne, F., Menard, I., Massonnat, G.J., Rolando, J.-P., 1999. Abnormal reverse faulting above a depleting reservoir. Geology 27, 111}114. Oliver, J., Johnson, T., Dorman, J., 1970. Postglacial faulting and seismicity in New York and Quebec. Canadian Journal of Earth Sciences 7, 579}590. Peltzer, G., Rosen, P., Rogez, F., Hudnut, K., 1996. Postseismic rebound in faulted step-overs caused by pore #uid #ow. Science 273, 1202}1207. Pollitz, F.F., Burgmann, R., Romanowicz, B., 1998. Viscosity of oceanic aesthenosphere inferred from remote triggering of earthquakes. Science 280, 1245}1249.
Rojstaczer, S.A., Wolf, S., 1992. Permeability changes associated with large earthquakes: an example from Loma Prieta, California, 10/17/89. Geology 20, 211}214. Rydelek, P.A., Sacks, I.S., 1999. Large earthquake occurrence a!ected by small stress changes. Bulletin of the Seismological Society of America 89, 822}828. Sample, J.C., Reid, M.R., 1998. Contrasting hydrogeologic regimes along strike-slip and thrust faults in the Oregon convergent margin: evidence from the chemistry of syntectonic carbonate cements and veins. Geological Society of America Bulletin 110, 48}59. Segall, P., 1989. Earthquakes triggered by #uid extraction. Geology 17, 942}946. Simpson, D.W., Leith, W.S., Scholtz, C.H., 1988. Two types of reservoir-induced seismicity. Bulletin of the Seismological Society of America 78, 2025}2040. Sissons, J.B., Cornish, R., 1982. Rapid localized glacioisostatic uplift at Glen Roy, Scotland. Nature 297, 213}214. Stewart, I.S., Hancock, P.I., 1994. Neotectonics. In: Hancock, P.L. (Ed.), Continental Deformation. Pergamon Press, Exeter, UK, pp. 370}409. Thorson, R.M., 1996. Earthquake recurrence and glacial loading in western Washington. Geological Society of America Bulletin 108, 1182}1191. Turcotte, D.L., Schubert, G., 1982. Geodynamics; Applications of Continuum Physics to Geological Problems. Wiley, New York, pp. 104}131. Van der Wateren, 1995. Processes of Glaciotectonism. In: Menzies, J. (Ed.), Modern Glacial Environments; Processes, Dynamics and Sediments. Butterworth Heinemann, Oxford, UK, pp. 309}335. Yeats, R.S., Sieh, K., Allen, C.R., 1997. The Geology of Earthquakes. Oxford Univ. Press, New York, pp. 17}58.
Glacial tectonics: a deeper perspective Robert M. Thorson* Department of Geology and Geophysics, University of Connecticut, 345 Mansxeld Road (U-45), Storrs, CT 06269, USA
Abstract The upper 5}10 km of the lithosphere is sensitive to slight changes ((0.1 MPa) in local stress caused by di!erential loading, #uid #ow, the mechanical transfer of strain between faults, and viscoelastic relaxation in the aesthenosphere. Lithospheric stresses induced by mass and #uid transfers associated with Quaternary ice sheets a!ected the tectonic regimes of stable cratons and active plate margins. In the latter case, it is di$cult to di!erentiate glacially induced fault displacements from nonglacial ones, particularly if residual glacial stresses are considered. Glaciotectonics, a sub-subdiscipline within Quaternary geology is historically focussed on reconstructing past glacier regimes and, by de"nition, does not include these e!ects. The term `glacial tectonicsa is hereby suggested for investigations focussed on the past and continuing in#uences of ice sheets on contemporary tectonics. 2000 Elsevier Science Ltd. All rights reserved.
1. Introduction Presently, there is a conceptual shift in the geosciences away from increasing specialization, towards more integrative problems at global scales, a shift embodied by the phrase `Earth System Sciencea (Kump et al., 1999). Simultaneously, technologically driven advances in instrumental and computational techniques are permitting earth scientists to identify and explain synoptic variations in topography, seismicity, and ambient crustal stress that were not recognizable a decade ago. The recent paper by Peltzer et al. (1996) is a case in point; they used synthetic aperture radar (SAR) interferometry to explain the disappearance of transient topographic anomalies at the centimeter scale resulting from the 1992 Landers (California, USA) earthquake, anomalies erased by the in-migration of #uids to 4 km depth. As these conceptual and technical trends continue, the seismotectonic e!ects of past #uctuations of Quaternary ice sheets are becoming easier to notice and explain. As a result, the role of glaciers in earth deformation is being recognized at a range of spatial scales: from deformation within the ice itself to the reactivation of tectonic faults in response to crustal unloading and viscoelastic relaxation of isostatic anomalies. Broader recognition of the multiple
* Tel.: 860-486-1396; fax: 860-486-1383. E-mail address: [email protected] (R.M. Thorson).
e!ects of glacial mass transfers is blurring the distinction between the study of tectonics, per se, and the study of `glaciotectonicsa, which, historically, has been primarily concerned with deformed glacial deposits. In this paper I show how the physical coupling between glaciation and crustal deformation extends far beyond the decollement between ice and its substrate (i.e. beyond the scope of glaciotectonics), and instead operates up to crustal scales. I begin by reviewing recent research illustrating the sensitivity of the earth's crust to stress di!erences far smaller than those associated with ice sheets. I then introduce examples of glacially induced, crustal scale deformation from a passive cratonic setting (Fennoscandia) and an active continental margin (Cascadian subduction zone). Next, I explore the basic mechanisms of glacially induced tectonics * e!ective vertical stress, membrane #exure, and traction at the base of the crust * and how each of these mechanisms is modi"ed by the contrasting tectonic domains associated with compressive, extensional, and transform strain. Finally, I revisit the question of the scope of glaciotectonics.
2. Sensitivity of the Earth's crust The ambient seismicity in many regions is extremely sensitive to small changes in stress, regardless of cause. For example: Rydelek and Sacks (1999) recently demonstrated that seemingly trivial changes in the con"ning
0277-3791/00/$ - see front matter 2000 Elsevier Science Ltd. All rights reserved. PII: S 0 2 7 7 - 3 7 9 1 ( 0 0 ) 0 0 0 6 8 - 8
1392
R.M. Thorson / Quaternary Science Reviews 19 (2000) 1391}1398
stress (0.1 MPa) are su$cient to nucleate earthquakes; King et al. (1994) proved that fault displacement in one locality can transfer stress to other faults at continental scales; and Pollitz et al. (1998) implicated the aesthenosphere in the transfer of stress across the North Paci"c Ocean from the Aleutian Islands to California. The role of crustal pore pressure in tectonic processes has also been recently clari"ed. For example: Rojstaczer and Wolf (1992) documented a tenfold increase in baseflow stream discharge following the 1989 Loma Prieta (California, M 7.1) over a radius up to 50 km from the epicenter; Episodic movements on great transform faults (Byerlee, 1993; Sample and Reid, 1998) are toggled by pore pressure variations; Thrust faults within accretionary wedges at continental margins (Bolton et al., 1999) are regulated by hydro-mechanical (valving) processes which partition an otherwise continuous process (subduction) into discrete episodes of deformation. Decollemonts would be mechanically impossible without high pore pressures. Arti"cial triggering of earthquakes by human activities, now well established, works primarily through the e!ects of #uid pressure. For example, the injection of petroleum brines, waste-water, and experimental tracers have initiated brittle faulting to a depth of 9 km (Jost et al., 1998). Conversely, the withdrawals of #uids in#uence seismicity to a minimum depth of 4 km (Gomberg and Wolf, 1999; Segall, 1989). The so-called `delayeda response of reservoir seismicity has been shown to be a consequence of the migration of #uid pressure towards potential asperities on faults, with rupture occurring when a threshold pressure is reached. Dry e!ects such as crustal loading by water-supply reservoirs (Simpson et al., 1988; Mandal et al., 1998), the extraction of mass from large open-pit mines, and underground explosions are generally less important than #uid e!ects. Stress changes caused by human activities are small with respect to those of ice sheets, present or past, which also precipitate and dampen earthquake processes. For example, Johnston (1987) calculated that the East Antarctic Ice Sheet was responsible for a loading stress reaching 59 MPa, although the change in e!ective stress is much lower. Glaciers, as agents of denudation, are also responsible for permanent mass transfers from highlands to depositional basins, but at a much slower rate than ice loads. The closest analogue to a glacially induced seismicity from #uids was the damaging Koyna (India) earthquake of December 10, 1967 the largest known reservoir earthquake (Mandal et al., 1998). Its main shock was a strike-slip event with a magnitude of 6.2 (Simpson et al., 1988) and a focal depth of less than 5 km, but aftershocks extended to at least 15 km depth in a zone nearly 10 km wide and 30 km long. Carefully instrumented variations in aftershock micro-seismicity indicate that the state of stress in the crust in Koyna was controlled by inter-annual variability between
wet and dry years as well as the water level in the reservoir. Earthquake nucleation processes similar to those beneath the Koyna reservoir * dry loading, poroelastic volume changes, #uid transfers * took place on a much larger scale in the vicinity of former Quaternary Ice Sheets. Although the direct loading and pore pressure e!ects are most easily understood, the growth and decay of ice sheets were also associated with crustal #exure and aesthenospheric relaxation. Although recognized as important for more than century, the tectonic coupling between ice sheets (the water component of the lithosphere) and the crust (the silicate component) is seldom considered in geologically based reviews of the seismotectonic literature (Yeats et al., 1997; Keller and Pinter, 1996).
3. Contrasting examples Postglacial fault scarps in northern Fennoscandia are the dramatic surface expressions of faults penetrating the Baltic shield to a depth of about 40 km (Arvidsson, 1996). The most conspicuous, the Parvie Fault in northern Sweden, with a vertical displacement of 10 m and a rupture length of 150 km (Lagerback, 1992; Muir-Wood, 1988) may have generated a deglacial earthquake with an extraordinary magnitude (Mw"8.2). Similar, albeit smaller, faults are known for the Canadian and Siberian shields (e.g. Dyke et al., 1991), in tectonic settings that requires the `externally imposed failure conditionsa (Johnston, 1996) caused by glacial loading. Although the Parvie Fault, as well as their counterparts on other shields, was formed during deglacial mass transfers, it remains somewhat active today, as evidenced by continuing micro-seismicity. Ironically, modern seismicity has a glacial heritage. Building on previous models (Hasegawa and Basham, 1989; James and Morgan, 1990; Johnston, 1987; MoK rner, 1980; Muir-Wood, 1988; Stewart and Hancock, 1994) I applied the idea of crustal-scale glaciotectonic deformation to an active plate margin in the Cascadia Subduction Zone, arguing that the tectonic processes in the fore-arc basin of the Puget Lowland were, and continue to be, compromised by the mass and #uid transfers associated with Quaternary glaciation (Thorson, 1996). More speci"cally, I predicted that a much simpler, preglacial ambient tectonic regime has become partitioned into four discrete time domains, all of which occur sequentially during each glacial cycle: E A phase of normal background stress in which the direct and delayed e!ects of glacial mass transfers are insigni"cant (normal Coulomb failure). E A phase of mass loading/crustal strengthening in which the combination of ice thickening and isostatic
R.M. Thorson / Quaternary Science Reviews 19 (2000) 1391}1398
out#ow reduced the deviatoric stresses, contributing to stability. E A phase of unloading/weakening during early postglacial time associated with signi"cant fault displacements caused by #uid overpressuring and enhanced traction at the base of the crust due to isostatic in#ow. E A recovery phase in which ambient tectonic stresses, supposedly `eraseda during the deglacial phase, reaccumulated to background values. An attempt to test this hypothesis with "eld data (Holocene fault displacements) was encouraging, but inconclusive. Since publication of that paper, the evidence for linking glacial unloading with a transient phase of intense deglacial tectonism has been strengthened. Nearly 40 years ago, Easterbrook (1963) published evidence for an enigmatic emergence and re-submergence of a crustal block near Deming, in northwestern Puget Lowland, during glaciomarine conditions about 12 ka. Owing to the magnitude of the required oscillation (30}70 m), the interpretation of these features remains controversial. Recently, Dragovich et al. (1997) linked these vertical motion anomalies (as well as a concentration of bedrock landslides) to the leading edge of the upper plate of a thrust fault exhibiting both Neogene o!set and modern seismic activity. Emergence was likely caused by at least 30 m of localized tectonic uplift during the most rapid phase of deglaciation, an interval when isostatic uplift took place at a rate between 10 and 30 cm yr\ (Dethier et al., 1995), and when the rapid northward #ow of a low-viscosity mantle was in phase with the plate convergence direction (10}10 poise; James et al., 2000). Tectonically, this uplift can be interpreted as a structural wedge caught between the leading edge of a Quaternary thrust fault and an antithetic reverse fault behind it. The subsequent re-submergence is more enigmatic, but may have been caused by normal faulting and backtilting associated with a subsequent phase of crustal extension associated with continued growth of the uplift dome (Muir-Wood, 2000). Curiously, two other elevated anomalies in uplift pro"le * Mt. Vernon (Dethier et al., 1995) and Alki Point (Thorson, 1996), also lie on the upper plate of thrusts (Johnson et al., 1999). The amplitudes of the topographic anomalies ('30 m for Deming on the McCauley Creek Thrust, ca. 22 m for Mt. Vernon on the Devil's Mountain Fault, and ca. 9 m for Strand A of the Seattle Fault) scale with fomer ice thickness and with the magnitude and rate of isostatic uplift.
4. Glacial tectonics The magnitude of the deglacial uplifts involved in Puget Lowland and the intimate connection between
1393
pre-existing tectonic structures and glacially induced deformation, blurs the distinction between tectonics and glacially induced tectonics. Strictly speaking, the glacier itself is a tectonic phenomenon. The base of each ice sheet is a rapidly propagating, gravity-driven detachment fault, taking place entirely within the lithosphere. The loss of ice by melting is analogous to un-roo"ng by denudation 2The glacier sole is essentially a slickensided decollement, smeared with fault gouge. The rigid lower plate (the crust) undergoes elastic #exure and changing deviatoric stresses caused directly by the load, and indirectly by the transient decay of the isostatic anomaly and excess pore #uid pressure. (Thorson, 1996, p. 1187) What separates glacial tectonics from `normala tectonics is less the physical processes involved or the scope of their in#uence, but the time scale.
5. Glacial tectonic mechanisms Kinematically, there are three principal directions in which the forces associated with a change in glacier mass can in#uence the state of stress of the lithosphere: The stresses can be imposed: vertically through changes in e!ective stress; horizontally by imparting a shear traction at the top of the crust (conventional glaciotectonics) or at its base (horizontal component of the viscoelastic response in the aesthenosphere); or three dimensionally with respect to volumetric strain and #exure. All of these e!ects, treated separately below, take place simultaneously, and can overlap in time (Fig. 1). 5.1. Ewective vertical stress Here we consider three limiting cases. In the simplest limiting case, pore pressure e!ects are absent, allowing the growth of an ice sheet (of in"nite width) to create a direct, instantaneous, lithostatic increase in the vertical stress: *p "o g h , T where *p is the change in vertical stress, o is the density of ice, and h is the ice thickness. This e!ect is equivalent to the `rapida reservoir response in induced seismicity directly beneath impoundments. Owing to the Poisson e!ect, however, the increased vertical stress also causes a proportional (but lower) increase in the horizontal stress, or *p , reducing the potential deviatoric (i.e. tec& tonic) stress. The importance of loading diminishes with depth as the stress ratio between pre and post-loaded conditions declines. The principal e!ects are elastic and instantaneous.
1394
R.M. Thorson / Quaternary Science Reviews 19 (2000) 1391}1398
Fig. 1. Schematic diagram showing glaciotectonic domains and processes. (a) Three fundamental processes associated with changes in glacier distribution; elastic e!ects, vertical stresses, and shear. (b) Idealized physical model for changes in vertical stress during loading. The change in e!ective stress is determined by changes in pore pressure (k) and vertical load (*p ), and are reversed and generally more rapid during unloading. Crustal 4 permeability (k) after Manning and Ingebritsen (1999). Change in vertical load with depth is idealized. Top part of the physical model has no scale; the thickness of layers is exaggerated to show di!erential responses.
In a second limiting case, a wet-based ice sheet is saturated with #uid water to near its surface, and the water has unlimited access to the base in an isotropic crust. In this circumstance, the change in e!ective vertical stress (*p ) is zero because the increased #uid pressure # (*k) fully compensates the increased vertical load (*p ). Alternatively *p "0"(*p !k). # This limiting case is realistic for ice-loaded shield areas well inside the glacial margin, where fractures in strong rocks to a depth of several kilometers are open to the surface, and where the high fracture permeability does not restrict the #ow of #uids. In these circumstances, changes in the e!ective vertical stresses remain close to zero because the #uid pressure instantaneously mirrors the changes in load. Changes in horizontal stresses through the Poisson e!ect remain important. In cases where the potentiometric surface of #uid water within the ice sheet falls to a level appreciably below the glacier surface, then the e!ective vertical stress is raised by the proportional decrease in #uid pressure. Conversely, during glacier surges, the potentiometric pressure from meltwater (o"1.0 g/cm) may exceed the added vertical load of the ice (o"0.9 g/cm), causing a transient reduction in the e!ective vertical stress during a time of rapid mass transfer. Thus, micro-seismicity associated
with glacier surges (Kamb et al., 1985) likely takes place in the bedrock, as well as within the ice. The permeability of the Earth's crust, whether sedimentary basins, fractured crystalline basement, and volcanic rocks, decreases systematically with depth, owing to increased con"ning pressure and the closure of fractures (Fig. 1). Based on studies of metamorphic #uids and heat transport (Manning and Ingebritsen, 1999), the permeability (k) in SK units of `ma varies with crustal depth (z, km) in a simple, log}log relationship spanning at least 10 orders of magnitude: Log k"!14}3.2 log z. Log k for sedimentary basins to a depth of 7 km ranges from !12 to !16 m\. Volcanic rocks and crystalline basement are typically one-to-two order of magnitudes less permeable. Although conceptually straightforward, de"ning the boundary between meteoric and connate water is di$cult in actual practice. The relationship between permeability and depth is least reliable in the shallow crust owing to greater variation in horizontal stress and to lithological heterogeneity. At some depth, fracture aperture is small enough to limit the permeability, forcing glacially induced pressure changes to take place gradually. The depth at which this transition takes place is not known, but likely ranges widely from about 0.5 km to more than several km. This time-dependent migration of #uids in the shallow crust
R.M. Thorson / Quaternary Science Reviews 19 (2000) 1391}1398
(Peltzer et al., 1996) is similar to the `delayeda e!ect for reservoir seismicity (Simpson et al., 1988) and injection experiments (Odonne et al., 1999), which takes place over time scales of weeks to years. In the glacially induced version of this scenario, changes in #uid pressure of `meteoric watera in the open, but low-permeability crust, will lag each instantaneous increment of load. This will cause an instantaneous rise in the e!ective vertical stress that will decay as the pressure re-equilibrates. Non-linear changes in the permeability may occur during re-equilibration, possibly through the geochemical e!ects (stresscorrosion) of in"ltrating water (Kisslinger, 1976). At greater depth, however, where poroelastic e!ects (Segall, 1989) predominate, connate #uid plays a critical role in supporting the crust. A glacially induced increase in vertical stress (*p ) will be fully compensated by an instan taneous rise in #uid pressure, one that will gradually bleed o! through #uid out-migration. This will cause a transient phase during which an initially weakened crust will strengthen, an e!ect opposite that in the next-higher layer, where the initially strengthened crust will weaken. At even greater depths, #uids are not free to migrate upward, the proportional change in hydrostatic pressure diminishes, and anisotropy plays a critical, and largely unknown, role in modulating the e!ects of the ice load. Near the edge of the glacier, however, the e!ects will be more complicated. Disregarding #uid e!ects, the di!erential weighting on either side of a glacier margin will be felt immediately as the crust is elastically compressed below the load, with no compression beyond it. When #uid e!ects are considered, the di!erential stresses are greater because the glacier imparts a new hydraulic gradient on the system. Seepage pressures, particularly near the ice margin, may also in#uence the ambient stress. Membrane #exure of the crust during isostatic compensation will cause radial compression and extension beneath the ice load and in the forebulge domain, respectively, deformation that will a!ect crustal permeability. 5.2. Horizontal traction; top and bottom Drag forces at the glacier}bed interface are controlled by the e!ective overburden stress (ice thickness reduced by #uid pressure), the viscoelastic properties of the ice, and the frictional properties (Mohr}Coulomb) of the boundary. Three limiting cases can be visualized. In the "rst case, thick ice is frozen to a #at, crystalline bed; horizontally #owing ice and the stationary bed form an ideal simple shear couple limited only by the slow rate of ice deformation; in this case, the shear on the rock is trivial. In the second case, the ice moves continuously above a very weak, deforming bed, which fully compensates the shear stresses. As before, negligible shear is propagated downward to the lithosphere. In the third case the ice is frozen (or frictionally bonded) to its bed near the margin, with groundwater out#ow taking place beneath the ice}bed
1395
interface in strata where the permeability is limiting. In this circumstance, the frictional bond between the glacier and its bed (whether rock or unconsolidated sediment) is stronger than in deeper materials, even that of jointed bedrock. The traction associated with the advancing ice can thus be transferred downward into the crust, with displacement along faults, both brittle or ductile, taking place to a depth of up to several hundred meters, and involving intact slabs of the lithosphere at the kilometer scale (Aber, 1988). The depth and behavior of such e!ects are controlled primarily by the strain rates being applied. Thrusting at the ice margin, which leads to the familiar push moraines and imbricate ridges, is conceptually similar to the behavior at accretionary wedges of subduction zones. A second type of horizonally}directed shear takes place at the base of the rigid crust through the e!ect of horizontal #ow within the asthenosphere (James and Morgan, 1990). Traditionally, studies of glacioisostasy have focussed on the vertical component of motion and on crustal #exure (MoK rner, 1980), but the horizontal #ow between the forebulge and the subglacial depression, can be quite rapid, up to a meter per year. Recent geodetic investigations in the western US con"rm that viscoelastic relaxation in the aesthenosphere after major earthquakes (Rydelek and Sacks, 1999) raises the strain along other, distant faults. Similarly, horizontal isostatic motions during loading and unloading must enhance or diminish the horizontal components of regional stress (Muir-Wood, 2000). 5.3. Three-dimensional yexure Mechanical loading is has been reviewed by Turcotte and Schubert (1982). Slab loads, line loads, point loads, end loads, and embedded loads all create #exures of the lithosphere, the size and shape of which are governed largely by the e!ective rigidity of the crust, which, is sensitive to the third power of membrane thickness. The physics of the process are independent of the composition of the load, whether glacier ice, crystallized lava, or a sedimentary accumulation. Hence, the familiar `moatforebulgea pro"les of volcanic islands, foreland basins, and glacial margins are similar. Importantly, membrane #exure includes zones of convexity where the crust is radially dilated, and zones of concavity where it is radially compressed. Radial and tangential extension on the forebulge, and radial compression beneath the ice sheets took place during the loading hemicycle; the reverse took place, and with higher strain rates, during deglaciation, a process that continues today. 6. Complications 6.1. Ambient crustal stress In the previous discussion, I made the simplifying assumption that the ambient lithospheric stress was
1396
R.M. Thorson / Quaternary Science Reviews 19 (2000) 1391}1398
isotropic. Lithospheric stress, however, varies continuously and heterogeneously in response to plate motion, topographic loading, exhumation, and the mass transfer of water during ice ages. Generally speaking, the state of crustal stress can be classi"ed into one of three categories: In a compressive, or thrust regime, the principal stress (p ) is horizontal and the minimum stress (p ) is vertical. Increased e!ective vertical stresses stabilizes the system by reducing the deviatoric stress. Conversely, a reduction in e!ective vertical stress pushes the system towards instability. Fluid injection often produces earthquakes with focal solutions consistent with at least a component of reverse faulting (Jost et al., 1998) because the e!ect of #uid injection is to raise the pore pressure, which reduces the e!ective vertical stress. In an extensional, normal fault regime, p is vertical and p is horizontal. Increas ing the e!ective vertical stress helps destabilize the system, whereas reducing the vertical stress stabilizes it. Fluid extraction (Gomberg and Wolf, 1999) increases vertical stress by reducing the pore pressure, and is thus often associated with normal faulting (Amelung et al., 1999). In a strike slip regime, where p and p are both horizontal, an increase in the e!ective stress stabilizes the system through the Poisson e!ect; pore pressure is especially important in regulating strike slip faults. Although there are broad tectonic domains where the ambient stress is relatively uniform, local stresses are often highly heterogeneous at a variety of scales (e.g. Mueller et al., 1999). Additionally, most faults have an oblique component of motion involving two of the idealized regimes above. 6.2. Glacial history The tectonic e!ects of Quaternary ice sheets are unique because their e!ective loads can change rapidly, more than several orders of magnitude faster than the processes of mountain building and erosion. Although the stresses are lower than for lithospheric forces, the stress gradients and strain rates can be signi"cantly higher, causing brittle rupture where ductile failure might have otherwise occurred. Also, unloading is usually much more rapid than loading, reversing many of the e!ects described above. Importantly, the e!ects of loading and unloading work in opposite directions, allowing the possibility for glaciers to trigger episodes of rupture regardless of ambient stress. They push and pull. 6.3. Coupled ewects Given the three basic directions of applied stress, the three fundamental stress domains, the heterogeneity of local stress within the crust at a variety of scales, and the asymmetric history of ice sheet loading and unloading, glaciotectonic stresses will be complex. Additionally, the e!ects can either amplify or diminish one another. For example:
E The creation of a forebulge may contribute to radial jointing, which may increase crustal permeability, which may decrease the pore pressure beneath the ice margin, which may steepen the ice pro"le, which may cause even greater #exure, etc. E Aesthenospheric out#ow caused by ice loading may enhance or diminish stresses of tectonic origin, depending on the ambient direction of plate convergence. For example, deglaciation of the Cordilleran Ice Sheet near Kodiak Island in southern Alaska may have enhanced and reduced the rate of plate convergence on the southern and northern #ank of the former ice sheet, respectively. E The rate of de-weighting caused by ablation may exceed the rate at which excess pore pressure can drain towards the surface, leaving the crust weaker than it would be under normal conditions. In thrust and strike-slip domains, this may lead to a deglacial interval of heightened seismicity. E Fault displacement triggered by a rise in pore pressure may be a self-limiting phenomenon through the e!ect of stress dilatancy, in which crustal porosity is increased through microfracturing. E Above the onset of poroelastic e!ects the crust (excluding normal fault regimes) is instantaneously strengthened by vertical loading, then weakened back to initial conditions by the in-migration of #uids. The opposite happens where poroelastic e!ects predominate; the crust is weakened, then strengthened. Both of these e!ects can take place simultaneously in di!erent superimposed crustal layers, concentrating stress in subhorizontal boundaries between these zones. E The permeability of the shallow crust is highly variable, hence the e!ects of changing glacial loads will be felt di!erentially; delayed pressure e!ects will continue to in#uence local stresses until re-equilibration, long after the ice has disappeared.
7. Conclusion The tectonic setting of passive continental margins is usually characterized by a thick granitic crust, greater uniformity in ambient stress, and low strain rates (10\ yr\). In such a setting * for example in Fennoscandia (Arvidsson, 1996), the British Isles (Sissons and Cornish, 1982), and the North American Craton (Adams and Bell, 1991; Andrews, 1991; Oliver et al., 1970) * glacially induced stresses are superimposed on the regional "eld, but may actually be the dominant factor controlling seismicity and crustal motion at the time scale of 10}10 yr (MoK rner, 1991). In contrast, the tectonic setting of active continental margins is characterized by high strain rates (10\ yr\), crustal-scale anisotropy, and the dynamic e!ects of plate #exure and buoyancy. In this type of setting * for example, in Cascadia (Mathews
R.M. Thorson / Quaternary Science Reviews 19 (2000) 1391}1398
et al., 1970; Clague, 1983; Thorson, 1996), glacially induced stresses and subduction work simultaneously at Quaternary time scales. In the former case of passive margins, ice sheets modify an existing tectonic regime. In the latter case, they must be considered part of it. Glaciotectonics, as conventionally de"ned, `2involves structural deformations of the upper horizon of the lithosphere caused by glacial stressesa (Van der Wateren, 1995). This de"nition speci"cally excludes deformation within the ice itself, and crustal-scale e!ects because the primary research objective of glaciotectonics, is the reconstruction of past glacier regimes from "eld observations, especially those associated with deformation of unconsolidated materials at and below the glacier sole, and with marginal thrusting at scales of 10\}10 m. The academic territory of glaciotectonics is clearly de"ned as subdiscipline within Quaternary geology. I recommend continued use of the term in this narrowly restricted sense. A new term, however is needed, one that encompasses the deformation in the shallow crust caused by ice}bed traction, and the crustal-scale deformation described in this review. For now I suggest the following candidate terms, arranged in order of increased simplicity: glacio-seismotectonics, glacio-seismicity, glacially induced tectonics, glacial tectonics, or simply tectonics. Although tempted to adopt the last candidate for its simplicity, the term `glacial tectonicsa seems an acceptable compromise between focus and breadth. The actual practice of glacial tectonics has a long pedigree. I recommend adoption of this term to help focus an important, coalescing subdiscipline within geology, one that straddles the boundary between glaciology and tectonics.
Acknowledgements Jeanne Sauber, Iain Stewart and Jim Rose are thanked for reviews that considereably improved the paper. Earlier discussions with Sam Johnson, Jean Crespi and Tom Torgersen were also helpful.
References Aber, J.S., 1988. Bibliography of glaciotectonic references. In: Croot, D.G. (Ed.), Glaciotectonics: Forms and Processes. A.A. Balkema, Rotterdam, pp. 195}210. Adams, J., Bell, J.S., 1991. Crustal stresses in Canada. In: Slemmons, D.B., Engdahl, E.R., Zoback, M.D., Blackwell, D.D. (Eds.), Neotectonics of North America. Geological Society of America Decade Map Volume, Boulder, Colorado, pp. 367}386. Amelung, F., Galloway, D.L., Bell, J.W., Zebker, H.A., Laczniak, R.J., 1999. Sensing the ups and downs of Las Vegas: InSAR reveals structural control of land subsidence and aquifer-system deformation. Geology 27, 483}486. Andrews, J.T., 1991. Late Quaternary glacial isostatic recovery of North America, Greenland, and Iceland: a neotectonics perspective.
1397
In: Slemmons, D.B., Engdahl, E.R., Zoback, M.D., Blackwell, D.D. (Eds.), Neotectonics of North America. Geological Society of America Decade Map Volume, Boulder, Colorado, pp. 473}486. Arvidsson, R., 1996. Fennoscandian earthquakes: Whole crustal rupturing related to postglacial rebound. Science 274, 744}746. Bolton, A.J., Clennell, M.B., Maltman, A.J., 1999. Nonlinear stress dependence of permeability: a mechanism for episodic #uid #ow in accretionary wedges. Geology 27, 239}242. Byerlee, J.D., 1993. Model for episodic #ow of high-pressure water in fault zones before earthquakes. Geology 21, 303}306. Clague, J.J., 1983. Glacio-isostatic e!ects of the Cordilleran ice sheet, British Colombia. In: Smith, D.E., Dawson, A.G. (Eds.), Shorelines and Isostasy. Academic Press, London, pp. 285}319. Dethier, D.P., Pessl Jr., F., Keuler, R.F., Balzarini, M.A., Pevear, D.R., 1995. Late Wisconsinan glaciomarine deposition and isostatic rebound, northern Puget Lowland, Washington. Geological Society of America Bulletin 107, 1288}1303. Dragovich., J.D., Zollweg, J.E., Qamar, A.I., Norman, D.K., 1997. The Macaulay Creek Thrust, the 1990 5.2-magnitude Deming Earthquake, and Quaternary geologic anomalies in the Deming Area, western Whatcom County, Washington * Cause and E!ects? Washington Geology 25, 15}26. Dyke, A.S., Morris, T.F., Green, D.E.C., 1991. Postglacial tectonic and sea level history of the central Canadian Arctic. Geological Survey of Canada Bulletin 397, 56 pp. Easterbrook, D.J., 1963. Late Pleistocene glacial events and relative sea-level changes in the northern Puget Lowland, Washington. Geological Society of America Bulletin 74, 1465}1483. Gomberg, J., Wolf, L., 1999. Possible cause for an improbable earthquake: the 1997 Mw 4.9 southern Alabama earthquake and hydrocarbon recovery. Geology 27, 367}370. Hasegawa, H.S., Basham, P.W., 1989. Spatial correlation between seismicity and postglacial rebound in eastern Canada. In: Gregersen, S., Basham, P.W. (Eds.), Earthquakes of North Atlantic Passive Margins. Kluwer Academic Publishers, Dordrecht, pp. 483}500. James, T.S., Clague, J.J., Wang, K., Hutchinson, I., 2000. Postglacial rebound at the northern Cascadia Subduction Zone. Journal of Quaternary Science Reviews, this issue. James, T.S., Morgan, W.J., 1990. Horizontal motions due to postglacial rebound: Geophysical Research Letters 17, 957}960. Johnson, S.Y., Dadisman, S.V., Childs, J.R., Stanley, W.D., 1999. Active tectonics of the Seattle Fault and central Puget Sound, Washington * implications for earthquake hazards. Geological Society of America Bulletin 111, 1042}1053. Johnston, A.C., 1987. Suppression of earthquakes by large continental ice sheets. Nature 3309, 467}469. Johnston, A.C., 1996. A wave in the earth. Science 274, 735}736. Jost, M.L., Bu{elberg, T., Jost, O., Harjes, H.-P., 1998. Source parameters of injection-induced microearthquakes at 9 km depth at the KTB deep drilling site, Germany. Bulletin of the Seismological Society of America 88, 815}832. Kamb, B., Raymond, C.F., Harrison, W.D., Engelhardt, H.F., Echelmeyer, K.A., Humphrey, N.F., Brigman, M.M., Pfe!er, T., 1985. Glacier surge mechanism: 1982}1983 surge of Variegated Glacier, Alaska. Science 227, 469}479. Keller, E.A., Pinter, N., 1996. Active Tectonics. Prentice-Hall, Upper Saddle River, NJ, 338 pp. King, G.C.P., Setin, R.S., Lin, J., 1994. Static stress changes and the triggering of earthquakes. Bulletin of the Seismological Society of America 84, 935}953. Kisslinger, C., 1976. A review of theories of mechanisms of induced seismicity. Engineering Geology 10, 85}98. Kump, L.R., Kasting, T.F., Crane, R.G., 1999. The Earth System. Prentice-Hall, Upper Saddle River NJ, 351 pp. Lagerback, R., 1992. Dating of Quaternary faulting in northern Sweden. Journal of the Geological Society of London 149, 285}292.
1398
R.M. Thorson / Quaternary Science Reviews 19 (2000) 1391}1398
Mandal, P., Rastogi, B.K., Sarma, C.S.P., 1998. Source parameters of Koyna earthquakes, India. Bulletin of the Seismological Society of America 88, 833}842. Manning, C.E., Ingebritsen, S.E., 1999. Permeability of the continental crust: implications of geothermal data and metamorphic systems. Reviews of Geophysics 37, 127}150. Mathews, W.H., Fyles, J.G., Nasmith, H.W., 1970. Postglacial crustal movements in southwestern British Columbia and adjacent Washington State. Canadian Journal of Earth Sciences 7, 690}702. MoK rner, N.-A., 1980. Earth Rheology, Isostasy and Eustasy. Wiley, New York, 599 pp. MoK rner, N.-A., 1991. Intense earthquakes and seismotectonics as a function of glacial Isostasy. Tectonophysics 188, 407}410. Mueller, K., Champion, J., Guccione, G., Kelson, K., 1999. Fault slip rates in the modern New Madrid Seismic Zone. Science 286, 1135}1138. Muir-Wood, R., 1988. Extraordinary deglaciation reverse faulting in northern Fennoscandia. In: Gregersen, S., Basham, P.W. (Eds.), Earthquakes at North Atlantic Passive Margins: Neotectonics and Postglacial Rebound. Kluwer Academic Publishers, Dordrecht, pp. 141}173. Muir-Wood, R., 2000. Deglaciation Seismotectonics: a principal in#uence on intraplate seismogenesis at high latitudes? Journal of Quaternary Science Reviews, this issue. Odonne, F., Menard, I., Massonnat, G.J., Rolando, J.-P., 1999. Abnormal reverse faulting above a depleting reservoir. Geology 27, 111}114. Oliver, J., Johnson, T., Dorman, J., 1970. Postglacial faulting and seismicity in New York and Quebec. Canadian Journal of Earth Sciences 7, 579}590. Peltzer, G., Rosen, P., Rogez, F., Hudnut, K., 1996. Postseismic rebound in faulted step-overs caused by pore #uid #ow. Science 273, 1202}1207. Pollitz, F.F., Burgmann, R., Romanowicz, B., 1998. Viscosity of oceanic aesthenosphere inferred from remote triggering of earthquakes. Science 280, 1245}1249.
Rojstaczer, S.A., Wolf, S., 1992. Permeability changes associated with large earthquakes: an example from Loma Prieta, California, 10/17/89. Geology 20, 211}214. Rydelek, P.A., Sacks, I.S., 1999. Large earthquake occurrence a!ected by small stress changes. Bulletin of the Seismological Society of America 89, 822}828. Sample, J.C., Reid, M.R., 1998. Contrasting hydrogeologic regimes along strike-slip and thrust faults in the Oregon convergent margin: evidence from the chemistry of syntectonic carbonate cements and veins. Geological Society of America Bulletin 110, 48}59. Segall, P., 1989. Earthquakes triggered by #uid extraction. Geology 17, 942}946. Simpson, D.W., Leith, W.S., Scholtz, C.H., 1988. Two types of reservoir-induced seismicity. Bulletin of the Seismological Society of America 78, 2025}2040. Sissons, J.B., Cornish, R., 1982. Rapid localized glacioisostatic uplift at Glen Roy, Scotland. Nature 297, 213}214. Stewart, I.S., Hancock, P.I., 1994. Neotectonics. In: Hancock, P.L. (Ed.), Continental Deformation. Pergamon Press, Exeter, UK, pp. 370}409. Thorson, R.M., 1996. Earthquake recurrence and glacial loading in western Washington. Geological Society of America Bulletin 108, 1182}1191. Turcotte, D.L., Schubert, G., 1982. Geodynamics; Applications of Continuum Physics to Geological Problems. Wiley, New York, pp. 104}131. Van der Wateren, 1995. Processes of Glaciotectonism. In: Menzies, J. (Ed.), Modern Glacial Environments; Processes, Dynamics and Sediments. Butterworth Heinemann, Oxford, UK, pp. 309}335. Yeats, R.S., Sieh, K., Allen, C.R., 1997. The Geology of Earthquakes. Oxford Univ. Press, New York, pp. 17}58.